Methods for Clear Call Under Noisy Conditions

1. A speech recovery system implemented with a noise cancellation apparatus comprising: a vibration sensor and a microphone for receiving and transmitting voice signals as incoming speeches from a speaker; and the vibration sensor and the noise cancellation apparatus further includes a speaker identity extracting module for extracting speaker identity representations from the incoming speeches as a referencing identity representation from the incoming speeches from the speaker; the vibration sensor further generates vibration signals corresponding to the voice signals without being affected by environmental noise signals for applying the vibration signals as reference signals for cancelling the environmental noise signals; wherein the speaker identity extracting module further applies signals from the vibration sensor and the microphone for extracting and applying the speaker identity representations into linguistic representations to synthesize clear full band high quality personal speech with same linguistic content and same speaker characteristics as the incoming speeches from the speaker; the noise cancellation apparatus further applies a SNR value to determine a value of quietness of acoustic environment form selecting proper vibration sensor and microphone sensor signals to convert the signals from one or a combination of both channels to a probabilistic distribution of linguistic (PDL) and converting the PDL and speaker identity representation to the speaker's microphone quality speech; and the noise cancellation apparatus further applies a SNR value to determine a value of quietness of acoustic environment for selecting proper vibration sensor and microphone sensor signals to convert the signals form one or a combination of both channels to the partial band PDL and the vibration sensor and a microphone sensor channel further applies a corresponding neural network model for converting a full band PDL, wherein a partial band PDL is converted from signals received from the vibration and the full band PDL is converted form signals from the microphone.

2. The speech recovery system implemented with a noise cancellation apparatus of claim 1 wherein: the vibration sensor includes a MEMS device comprising piezoelectric accelerometers for installation in earbuds as a wearable apparatus.

3. The speech recovery system implemented with a noise cancellation apparatus of claim 1 wherein: the vibration sensor includes a MEMS device comprising piezoelectric accelerometers for installation in necklaces as a wearable apparatus.

4. The speech recovery system implemented with a noise cancellation apparatus of claim 1 wherein: the vibration sensor includes a MEMS device comprising piezoelectric accelerometers for installation in patches directly on an upper body for detecting vibrations.

5. The speech recovery system implemented with a noise cancellation apparatus of claim 1 wherein: the vibration sensor further includes a laser-based vibration sensor for detecting and sensing a non-contact vibration.

6. The speech recovery system implemented with a noise cancellation apparatus of claim 1 further comprising: a wireless transmitter/receiver to transmit and receive wireless signals.

7. The speech recovery system implemented with a noise cancellation apparatus of claim 1 wherein: the speaker identity extracting module further comprises a converter for applying a rapid adapted conversion functional module to convert the voice signals and vibration signals to probabilistic distribution of linguistic representation sequences (PDLs) and wherein the noise cancellation apparatus further comprises a mapping module that is first developed and trained in an adaptation phase for mapping the PDL into a full band MCEP sequence.

8. The speech recovery system implemented with a noise cancellation apparatus of claim 7 wherein: the speaker identity extracting module further comprises a converter for applying a rapid adapted conversion model to convert the voice signals detected by the vibration sensor to probabilistic distribution of linguistic representation sequences (PDLs).

9. The speech recovery system implemented with a noise cancellation apparatus of claim 7 wherein: the speaker identity extracting module further comprises a converter for applying a rapidly adapted conversion model to convert the voice signals detected by the microphone to probabilistic distribution of linguistic representation sequences (PDL).

10. The speech recovery system implemented with a noise cancellation apparatus of claim 1 wherein: the speaker identity extracting module includes a plurality of brainwave sensors for mounting on a speaker's head for sensing signals from the speaker's head to convert the signals from the speaker head to the speaker identity representations.

11. The speech recovery system implemented with a noise cancellation apparatus of claim 1 wherein: the speaker identity extracting module is further implemented by applying a speaker embedding an i-victor process for mapping speaker independent linguistic representations to a speaker dependent mel-spetrogram.

12. The speech recovery system implemented with a noise cancellation apparatus of claim 1 wherein: the microphone further has a reduced volume channel for receiving and detecting background sounds to mix with a synthesized speaker voice based on a setting of a user's earbud when setting on a semi-transparent mode for reducing background sounds generated at a speaker side.

13. The speech recovery system implemented with a noise cancellation apparatus of claim 1 further comparing: adaptive rubber buds for implementing active noise cancellation algorithms for generating a background blocking mode to block background sounds receivable by other communication parties.

14. The speech recovery system implemented with a noise cancellation apparatus of claim 1 further comprising: a far-field automatic speech recognizer (FF-ASR) fort translating the incoming speeches from a speaker into PDL for converting to a clean voice of the speaker in various implementations of various live communication applications.

15. The speech recovery system implemented with a noise cancellation apparatus of claim 13 wherein: the FF-ASR further implements various noise-cancellation techniques, including beamforming, reverberation removal, coupled with multiple speech channels or recordings via microphone arrays.